Chinese short dramas live and die by their lines — fast, slangy, full of idioms that carry the joke or the insult. And those are exactly the lines that break in translation.
Take this one (from a fictional demo script I wrote for testing):
林晓：你少在这儿装蒜了，你那点小心思，我一眼就看穿了。
"Stop playing dumb — I can see right through your little scheme."
装蒜 (zhuāng suàn) literally involves 蒜, garlic. A word-for-word Spanish rendering gives you hacerse el ajo — "to make yourself the garlic" — which sends viewers picturing produce instead of following the argument.
A native-speaking dubbing editor would never translate the image. They'd translate the function: "No te hagas el tonto conmigo" in Spanish, "Não se faça de bobo comigo" in Brazilian Portuguese — both are the local fixed expressions for "playing dumb." Same punch, no garlic.
That gap — between what a translator (human or AI) produces on the first pass and what an editor would flag — is what I wanted to automate. Not the translation itself. The check.
The three ways lines actually break
Working through test lines, the failures clustered into three buckets:
- Idiom images. 手头紧 ("tight on money") literally says "my hand is tight." Spanish just says estar corto de dinero. The image must be dropped, not carried over.
- Tone carried by form, not words. 行行行，你说的都对，我错了还不行吗？ is a sarcastic non-apology. Its literal translation reads sincere. The fix in Spanish is rhythmic: Ya, ya, ya… tú siempre tienes la razón. ¿Contento? — the triple repetition does the eye-roll.
- Address terms. Chinese uses kinship words as polite address （阿姨 for a landlady). Tía in Spanish means your actual aunt. The functional equivalents are señora (ES) and dona + name (PT-BR).
A single "translate this naturally" prompt occasionally lands these. Occasionally is the problem — with 60 lines an episode, you need a fixed procedure that flags every line the same way.
Why a Skill file
I packaged the check as a Claude Skill (SKILL.md) instead of a saved prompt, for one reason: a Skill carries a workflow and an output contract, not just an instruction. The structure:
- Trigger conditions — paste Chinese lines (or an existing ES/PT draft to be diagnosed), get the check; it doesn't fire on general translation chat.
- A 5-step pass per line — read for intent → flag literal-translation stiffness → rewrite in neutral Latin American Spanish → rewrite in Brazilian Portuguese → handle puns/culture items explicitly.
- A fixed output format — original line → problem flags → ES rewrite + note → PT-BR rewrite + note → pun/culture workaround when needed → batch summary. The notes are in Chinese, so the creator can judge the edit without reading the target language.
- A fully worked example inside the file — so the model anchors on a real input→output pair, not just rules.
One rule I wrote in bold, in my head at least: flag uncertain expressions for native review instead of guessing. The skill is a first pass. Anything broadcast still gets human native eyes at the end — the file says so, and so does the product page.
What a run looks like
For the sarcasm line above, the output flags that a literal rendering "只剩字面认错，反讽语气全丢" (keeps the literal apology, loses all the sarcasm), then gives:
- ES: Ya, ya, ya… tú siempre tienes la razón. ¿Contento?
- PT-BR: Tá, tá, tá… você tá sempre certo. Satisfeito?
- Note: the "行行行" effect is a repetition effect — the target language must reproduce the rhythm with its own repeated filler (ya / tá), not collapse it into a single "okay."
Six demo lines (all fictional — slang, address terms, an oath, sarcasm) are included in the pack so you can see the full format before deciding if the workflow fits yours.
The file
The skill, the worked demo, and the fixed output format are here for $19, one-time:
https://promptandskillco.gumroad.com/l/pidqzx
If you localize short-form video into Spanish or Portuguese, I'd genuinely like to know which line type breaks most often for you — idioms, tone, or address terms. That's the category the next version of the checker will grow into.
